Time to reappraise or distract? temporal and situational context in emotion regulation in daily life

Abstract

Contextual factors influence how people regulate their everyday emotions. While daily life is rich with situations that evoke emotion regulation, few studies have broadly investigated the role of context in regulating emotions in response to naturally occurring negative events. In this study, we use a structured diary technique—the Experience Sampling Method—to test how different types of contextual factors are associated with using reappraisal and distraction to regulate daily emotions in N = 74 young adults from the general population. The following contextual factors were assessed: time of the day, weekday, tiredness, event stressfulness, and event type. We found that higher stressfulness of negative events was associated with using more distraction within- and between-person and using more reappraisal between persons. Time of day and weekday were not associated with reappraisal or distraction use, suggesting that variation in people’s external environments due to temporal patterns does not influence reappraisal or distraction use. However, tiredness was positively associated with distraction and reappraisal use within persons. Exploratory analyses suggested that experiencing time pressure affords less distraction use, and that experiencing physical discomfort affords less reappraisal use. These findings underscore the dynamic nature of emotion regulation, and the importance of context in everyday emotion regulation.

Appendices

Appendix 1

The Twins dataset (Collip et al., 2013; De Wild-Hartmann et al., 2013) from the MERGE dataset library (for a list of studies included in the MERGE, see Appendix 1 in Rintala et al., 2019) included most of the variables also involved in the analyses of the current study. We, therefore, conducted a power analysis using the Twins dataset with variables: negative event intensity, tiredness, time of the day, weekday, and the covariates age and gender. Negative event intensity was measured as event unpleasantness ranging from 0 to -3 and tiredness was measured with one ESM item: I feel tired (1 = not at all to 7 = very much). Time of the day was transformed into the same nine time blocks as in the current study and weekday was included as a 7-level factor. The specific outcomes of reappraisal and distraction were not available in any existing dataset we had access to, therefore, as a proxy we used another emotion regulation variable available within the Twins dataset, rumination (I am ruminating, 1 = not at all—7 = very much). We considered rumination to be a satisfactory proxy for distraction and reappraisal because it provides an estimate of the expected occurrence of an emotion regulation process in daily life as well as its relationship with other daily life variables. Moreover, as hypothesized in the current investigation regarding reappraisal and distraction, rumination has been shown to vary along temporal patterns in daily life (Takano & Tanno, 2011).

The event type variable or a similar proxy, was not available in the Twins dataset. Therefore, we conducted a separate additional sensitivity analysis with estimates of the proportion of interpersonal events from another existing study within the MERGE library, the ZAPP study (Thewissen et al., 2008). In this study, participants were asked to think about the most important event since the last beep and to report whether the event was personal or social (0 = intrapersonal, 1 = interpersonal). Only events which were rated as negative were included. In this additional sensitivity analysis, to estimate power we varied the effect of the variable event type by allowing for an increase of 1%, 2%, 5%, 10%, 20%, 50%, and 100% in the fixed intercept. This means that when participants report an event as being interpersonal, there would be an increase in the fixed intercept of 1%, 2%,…, or 100%. This way, we gained an estimate of how power is affected by expected change in the effect of event type. Because the event type variable parameter estimates were derived from a different dataset, we did not have information about its random effects and therefore, we only estimated the fixed effect in the sensitivity analysis. We also assumed that the event type was linearly independent with the rest of the predictors included in the model, therefore, we used the parameter estimates from the power analysis reported above.

The power analysis was conducted with a Monte Carlo simulation with 1000 replicates with the parameter estimates derived from the Twins dataset and the ZAPP dataset. The study duration was set to 14 days with 3, 6, and 9 measurement occasions per day per participant condition, with 50 participants in each condition, as in the dataset that was used for the current study (current study, N = 74). The model for power analysis included rumination as the outcome variable, with time of the day, day of the week, event intensity, and tiredness as independent variables with random slopes and age as a covariate. Gender was not included as a covariate because there were no males in the Twins dataset used for power analysis. For each simulated sample, the power was estimated as the number of Monte Carlo replicates in which the null hypothesis that the fixed intercepts are not statistically significant from zero, is rejected. At the alpha level 0.05, the following estimates for power were observed for a sample size of 150: time of the day (block 4, 15:00 – 16:30, = 0.88 and block 8, 21:00 – 22:30 = 0.89), day of the week (weekday 5 = 0.93), tiredness (1.0), and event intensity (1.0). Age was not significantly associated with the outcome variable and therefore power to detect its effect was not calculated. The reappraisal, event type, and event stressfulness variables were only included in the long-format ESM questionnaire condition, in which we expected to have 78 participants. At the alpha level 0.05, the following estimates for power were observed for a sample size of 78: time of the day block 8, 21:00 – 22:30 = 0.62), day of the week (weekday 5 = 0.67), tiredness (0.97), and event intensity (1.0). Therefore, the power to detect the hypothesized effects of temporal variables time of day and weekday was potentially low (Hypotheses 1a and 1b). The sensitivity analyses showed that when the event type variable (coded as 0 = personal event, 1 = social event) explained an increase of 5% of in the intercept of the outcome variable, the power to detect the effect was 0.99.

We recognize that the parameter estimates used in this power analysis are not derived from a dataset that is identical to the dataset of the current study and we therefore do not assume this power analysis to be precise, but rather an estimate of how these models fit real data where we expect the effects to be similar to those in the current study. We hope therefore to provide a valuable power estimation derived from real ESM data that closely resembles the dataset used for the current study.

Appendix 2

To access the dataset, we submitted an abstract via the Data Curation for Open Science (DROPS; Kirtley, 2022)—data access system of the Center for Contextual Psychiatry, KU Leuven, which was reviewed by the data manager of the dataset of the original study. After the abstract was approved, we then submitted a request for the variables used for the proposed analyses for the current study. As a part of the variable request, we provided a link to an OSF registration. Variables were then released to the lead researcher, along with a time- and date-stamped receipt of data access. The ESM items used in the current study are also publicly available to other researchers at the ESM Item Repository, and online ongoing project where ESM items, details of their use, and development are documented (Kirtley et al., 2023; https://esmitemrepository.com/).

Appendix 3

Table 5

Table 5 Distraction and reappraisal use in daily life. Results of the effects of event types in the mixed multilevel analyses. Effect size β per independent variable and per emotion regulation outcome (N = 74)